On Aug 16, 5:19=A0pm, John Larkin <jlar...@highlandtechnology.com> wrote:> On Thu, 16 Aug 2012 12:20:49 -0700 (PDT), dagmargoodb...@yahoo.com > wrote:> >I invented loading LM324 outputs all by myself waayyyy back as a punk > >kid, to stiffen them up and kill the crossover THD. =A0Of course > >everyone else on the planet thought of it too. :-) > > Me too! And I also invented loading the charge pump phase detector in > the 4046. > > I also invented the successive-detection log video detector, and the > dual-slope ADC. I did those at times when I could have patented them. > I'd be smoking seegars on the Riviera.Rivieras get boring, and seegars are bad for you--you came out better! -- Cheers, James Arthur
fast ramp follies
Started by ●August 14, 2012
Reply by ●August 16, 20122012-08-16
Reply by ●August 16, 20122012-08-16
On Thu, 16 Aug 2012 15:05:17 -0700 (PDT), dagmargoodboat@yahoo.com wrote:>On Aug 16, 2:57�pm, whit3rd <whit...@gmail.com> wrote: >> On Wednesday, August 15, 2012 1:17:18 PM UTC-7, (unknown) wrote: >> > On Aug 15, 2:58�pm, whit3rd <whit...@gmail.com> wrote: >> >> > > On Monday, August 13, 2012 8:49:33 PM UTC-7, John Larkin wrote: >> >> > > > On Mon, 13 Aug 2012 20:38:56 -0700, miso <m...@sushi.com> wrote: >> >> > > > >> OK, I need to charge a capacitor with a stable constant current. The >> >> > > > >> desired slope is about a volt per nanosecond. >> >> > > > >I'd be more inclined to cascode the current source output rather than >> >> > > > >add an inductor. >> >> > > > Would that help? The beta error would increase, and cascodes can >> >> > > > oscillate >> >> > > If beta 'error' is a problem,... >> > > you can add a transistor base resistor and feed back to the op amp >> >> > > to completely remove base current >> > Is this the one you're thinking of? >> >> >http://groups.google.com/group/sci.electronics.design/browse_thread/t... >> >> Looks like the same thing; BAD ASCII ART FOLLOWS >> >> � � � � � � �+---- 47k-----+ � � � � � � (Iout) >> � � � � � � �| � � � � � � | � � � � � � � � | >> � � � � � � �| � |\ � � � �| � � � � � � � � | >> Vref>--1k----+---| \ � � � | � � � � � � � |/ >> � � � � � � � � �| �>---+--+---47ohm----+--| >> � � � � � � �+---|-/ � �| � � � � � � � | �| >> � � � � � � �| � |/ � � = � � � � � � � | �|\ >> � � � � � � �+----------+---+--47k------+ � �V >> � � � � � � � � � � � � � � | � � � � � � � �| >> � � � � � � � � � � � � � � +--1k------------+ >> � � � � � � � � � � � � � � � � � � � � � � �| >> � � � � � � � � � � � � � � � � � � � � � �(1ohm) >> � � � � � � � � � � � � � � � � � � � � � � �| >> � � � � � � � � � � � � � � � � � � � � � � GND >> >> The use of a cascode can be folded into this scheme, by (1) using matched >> transistors for the cascode (2) doubling the feedback (to compensate two >> base currents, while only sensing one). � A 1.2V reference as well as 2.5V cascode >> base bias can be had from a LV431 style programmable zener; set it for 2.5V >> and the sense node gives you a stable 1.25V.. > >Yes. As cute as i(b) cancellation is, Phil's pseudo-darlington has >two advantages-- > 1. by reducing i(b), the i(b) error becomes less important, and > 2. it should be more linear, since it removes Early's effect.The OpAmp FB pretty much takes out Early effect. So your left with making it transient-capable. ...Jim Thompson -- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food.
Reply by ●August 16, 20122012-08-16
On Aug 16, 6:41=A0pm, Jim Thompson <To-Email-Use-The-Envelope-I...@On-My- Web-Site.com> wrote:> On Thu, 16 Aug 2012 15:05:17 -0700 (PDT), dagmargoodb...@yahoo.com > wrote: > > > > > > > > > > >On Aug 16, 2:57 pm, whit3rd <whit...@gmail.com> wrote: > >> On Wednesday, August 15, 2012 1:17:18 PM UTC-7, (unknown) wrote: > >> > On Aug 15, 2:58 pm, whit3rd <whit...@gmail.com> wrote: > > >> > > On Monday, August 13, 2012 8:49:33 PM UTC-7, John Larkin wrote: > > >> > > > On Mon, 13 Aug 2012 20:38:56 -0700, miso <m...@sushi.com> wrote: > > >> > > > >> OK, I need to charge a capacitor with a stable constant curre=nt. The> > >> > > > >> desired slope is about a volt per nanosecond. > > >> > > > >I'd be more inclined to cascode the current source output rathe=r than> > >> > > > >add an inductor. > > >> > > > Would that help? The beta error would increase, and cascodes can > > >> > > > oscillate > > >> > > If beta 'error' is a problem,... > >> > > you can add a transistor base resistor and feed back to the op amp > > >> > > to completely remove base current > >> > Is this the one you're thinking of? > > >> >http://groups.google.com/group/sci.electronics.design/browse_thread/t=...> > >> Looks like the same thing; BAD ASCII ART FOLLOWS > > >> +---- 47k-----+ (Iout) > >> | | | > >> | |\ | | > >> Vref>--1k----+---| \ | |/ > >> | >---+--+---47ohm----+--| > >> +---|-/ | | | > >> | |/ =3D | |\ > >> +----------+---+--47k------+ V > >> | | > >> +--1k------------+ > >> | > >> (1ohm) > >> | > >> GND > > >> The use of a cascode can be folded into this scheme, by (1) using matc=hed> >> transistors for the cascode (2) doubling the feedback (to compensate t=wo> >> base currents, while only sensing one). A 1.2V reference as well as 2.=5V cascode> >> base bias can be had from a LV431 style programmable zener; set it for=2.5V> >> and the sense node gives you a stable 1.25V.. > > >Yes. =A0As cute as i(b) cancellation is, Phil's pseudo-darlington has > >two advantages-- > > 1. by reducing i(b), the i(b) error becomes less important, and > > 2. it should be more linear, since it removes Early's effect. > > The OpAmp FB pretty much takes out Early effect. =A0So your left with > making it transient-capable.But only for d.c., right?--the op amp compensation doesn't work at 1V/ nS. -- Cheers, James Arthur
Reply by ●August 16, 20122012-08-16
On Thu, 16 Aug 2012 16:08:13 -0700 (PDT), dagmargoodboat@yahoo.com wrote:>On Aug 16, 6:41�pm, Jim Thompson <To-Email-Use-The-Envelope-I...@On-My- >Web-Site.com> wrote: >> On Thu, 16 Aug 2012 15:05:17 -0700 (PDT), dagmargoodb...@yahoo.com >> wrote: >> >> >> >> >> >> >> >> >> >> >On Aug 16, 2:57 pm, whit3rd <whit...@gmail.com> wrote: >> >> On Wednesday, August 15, 2012 1:17:18 PM UTC-7, (unknown) wrote: >> >> > On Aug 15, 2:58 pm, whit3rd <whit...@gmail.com> wrote: >> >> >> > > On Monday, August 13, 2012 8:49:33 PM UTC-7, John Larkin wrote: >> >> >> > > > On Mon, 13 Aug 2012 20:38:56 -0700, miso <m...@sushi.com> wrote: >> >> >> > > > >> OK, I need to charge a capacitor with a stable constant current. The >> >> >> > > > >> desired slope is about a volt per nanosecond. >> >> >> > > > >I'd be more inclined to cascode the current source output rather than >> >> >> > > > >add an inductor. >> >> >> > > > Would that help? The beta error would increase, and cascodes can >> >> >> > > > oscillate >> >> >> > > If beta 'error' is a problem,... >> >> > > you can add a transistor base resistor and feed back to the op amp >> >> >> > > to completely remove base current >> >> > Is this the one you're thinking of? >> >> >> >http://groups.google.com/group/sci.electronics.design/browse_thread/t... >> >> >> Looks like the same thing; BAD ASCII ART FOLLOWS >> >> >> +---- 47k-----+ (Iout) >> >> | | | >> >> | |\ | | >> >> Vref>--1k----+---| \ | |/ >> >> | >---+--+---47ohm----+--| >> >> +---|-/ | | | >> >> | |/ = | |\ >> >> +----------+---+--47k------+ V >> >> | | >> >> +--1k------------+ >> >> | >> >> (1ohm) >> >> | >> >> GND >> >> >> The use of a cascode can be folded into this scheme, by (1) using matched >> >> transistors for the cascode (2) doubling the feedback (to compensate two >> >> base currents, while only sensing one). A 1.2V reference as well as 2.5V cascode >> >> base bias can be had from a LV431 style programmable zener; set it for 2.5V >> >> and the sense node gives you a stable 1.25V.. >> >> >Yes. �As cute as i(b) cancellation is, Phil's pseudo-darlington has >> >two advantages-- >> > 1. by reducing i(b), the i(b) error becomes less important, and >> > 2. it should be more linear, since it removes Early's effect. >> >> The OpAmp FB pretty much takes out Early effect. �So your left with >> making it transient-capable. > >But only for d.c., right?--the op amp compensation doesn't work at 1V/ >nS.Yep. That's why I misspelled and said, "So you're left with making it transient-capable." ;-) ...Jim Thompson -- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food.
Reply by ●August 16, 20122012-08-16
On Aug 16, 7:18=A0pm, Jim Thompson <To-Email-Use-The-Envelope-I...@On-My- Web-Site.com> wrote:> On Thu, 16 Aug 2012 16:08:13 -0700 (PDT), dagmargoodb...@yahoo.com > wrote: > > > > > > > > > > >On Aug 16, 6:41=A0pm, Jim Thompson <To-Email-Use-The-Envelope-I...@On-My=-> >Web-Site.com> wrote: > >> On Thu, 16 Aug 2012 15:05:17 -0700 (PDT), dagmargoodb...@yahoo.com > >> wrote: > > >> >On Aug 16, 2:57 pm, whit3rd <whit...@gmail.com> wrote: > >> >> On Wednesday, August 15, 2012 1:17:18 PM UTC-7, (unknown) wrote: > >> >> > On Aug 15, 2:58 pm, whit3rd <whit...@gmail.com> wrote: > > >> >> > > On Monday, August 13, 2012 8:49:33 PM UTC-7, John Larkin wrote: > > >> >> > > > On Mon, 13 Aug 2012 20:38:56 -0700, miso <m...@sushi.com> wro=te:> > >> >> > > > >> OK, I need to charge a capacitor with a stable constant cu=rrent. The> > >> >> > > > >> desired slope is about a volt per nanosecond. > > >> >> > > > >I'd be more inclined to cascode the current source output ra=ther than> > >> >> > > > >add an inductor. > > >> >> > > > Would that help? The beta error would increase, and cascodes =can> > >> >> > > > oscillate > > >> >> > > If beta 'error' is a problem,... > >> >> > > you can add a transistor base resistor and feed back to the op =amp> > >> >> > > to completely remove base current > >> >> > Is this the one you're thinking of? > > >> >> >http://groups.google.com/group/sci.electronics.design/browse_threa=d/t...> > >> >> Looks like the same thing; BAD ASCII ART FOLLOWS > > >> >> +---- 47k-----+ (Iout) > >> >> | | | > >> >> | |\ | | > >> >> Vref>--1k----+---| \ | |/ > >> >> | >---+--+---47ohm----+--| > >> >> +---|-/ | | | > >> >> | |/ =3D | |\ > >> >> +----------+---+--47k------+ V > >> >> | | > >> >> +--1k------------+ > >> >> | > >> >> (1ohm) > >> >> | > >> >> GND > > >> >> The use of a cascode can be folded into this scheme, by (1) using m=atched> >> >> transistors for the cascode (2) doubling the feedback (to compensat=e two> >> >> base currents, while only sensing one). A 1.2V reference as well as=2.5V cascode> >> >> base bias can be had from a LV431 style programmable zener; set it =for 2.5V> >> >> and the sense node gives you a stable 1.25V.. > > >> >Yes. =A0As cute as i(b) cancellation is, Phil's pseudo-darlington has > >> >two advantages-- > >> > 1. by reducing i(b), the i(b) error becomes less important, and > >> > 2. it should be more linear, since it removes Early's effect. > > >> The OpAmp FB pretty much takes out Early effect. =A0So you're left wit=h> >> making it transient-capable. > > >But only for d.c., right?--the op amp compensation doesn't work at 1V/ > >nS. > > Yep. =A0That's why I misspelledMagically corrected. :-)> and said, "So you're left with making it > transient-capable." ;-)Oh, but there are two cases: the op amp servo, and the transistor CCS step response. The former matters long-term, but doesn't fix the CCS' Early non- linearity as it ramps 1V/nS. Phil made that point a ways back. -- Cheers, James Arthur
Reply by ●August 16, 20122012-08-16
On Thu, 16 Aug 2012 15:08:21 -0700 (PDT), dagmargoodboat@yahoo.com wrote:>On Aug 16, 5:15�pm, John Larkin <jlar...@highlandtechnology.com> >wrote: >> On Thu, 16 Aug 2012 13:12:50 -0700 (PDT), dagmargoodb...@yahoo.com >> wrote: >> >> >> >> >> >> >> >> >> >> >On Aug 16, 3:59�pm, Jim Thompson <To-Email-Use-The-Envelope-I...@On-My- >> >Web-Site.com> wrote: >> >> On Thu, 16 Aug 2012 12:24:51 -0700 (PDT), dagmargoodb...@yahoo.com >> >> wrote: >> >> >> >On Aug 16, 11:41 am, John Larkin >> >> ><jjlar...@highNOTlandTHIStechnologyPART.com> wrote: >> >> >> On Thu, 16 Aug 2012 08:29:46 -0700, Jim Thompson >> >> >> >> <To-Email-Use-The-Envelope-I...@On-My-Web-Site.com> wrote: >> >> >> >On Thu, 16 Aug 2012 07:39:04 -0700 (PDT), dagmargoodb...@yahoo.com >> >> >> >wrote: >> >> >> >> >>On Aug 15, 10:43 pm, John Larkin >> >> >> >><jjlar...@highNOTlandTHIStechnologyPART.com> wrote: >> >> >> >> >>> Hey, how about this? >> >> >> >> >>>https://dl.dropbox.com/u/53724080/Circuits/Isrc_simpler.JPG >> >> >> >> >>It's kind of a variation on Phil's active inductor. That defeats all >> >> >> >>the oscillation modes I can think of so far. (c-b, e-b, c-e)(Which >> >> >> >>means little, since circuits are so ingenious inventing new ones.) >> >> >> >> >>> Ludicrously simple. No Miller nonsense. Might not oscillate. MMBTH81 >> >> >> >>> would have about 2 pF out, or BFT92 would have about 0.75 but lower >> >> >> >>> beta. >> >> >> >> >>You could use the beta-cancellation trick. But, stray and parasitic >> >> >> >>capacitance will be big factors compared to a 10pF integration cap. >> >> >> >> >>> If this works, I should have done it years ago. >> >> >> >> >Yup. But I'd still add a base resistor, for the very reasons Gerhard >> >> >> >Hoffmann brought up... the output Z of the OpAmp is a big unknown (*). >> >> >> >> The base and emitter are shorted at AC. The opamp Zout doesn't matter >> >> >> any more. >> >> >> >Yep, that's a good point. �If the op amp can drive that load all by >> >> >itself, it seems okay. �That load being 1nF in series with 120ohms, it >> >> >should be a snap. >> >> >> >> >(*) Which brings to mind... maybe add some DC load to the OpAmp to >> >> >> >keep it out of the class-B region. >> >> >> >> LM7301 is a r-r output design. It isn't an ancient class-B like an >> >> >> LM324. >> >> >> >I love LM324As. �Yeah, I know you hate 'em. >> >> >> CMOS R-R OpAmps inherently have load-dependent gain and phase, so I >> >> avoid them when I can. >> >> >But John rightly points out that in the ramp timescale, Vbe is held >> >constant by the 1nF cap. �So, the transistor's fine. >> >> >In the longer term a load-step on the op amp's output may make it >> >overshoot or ring. �That would happen opening and closing the shorting >> >switch. �Probably not a big deal, AFAICT. >> >> >I was initially concerned the op amp is effectively absent on the 1nS >> >response level, so there's no base drive, but John's right, that >> >behavior is set by the 1nF b-e cap. �IOW, it doesn't matter. >> >> The opamp is slowly servoing Vbe to get the average current right. The >> ramp is so fast that the transistor is still running in constant Vbe >> mode during the ramp. The cap could actually be much bigger... doesn't >> matter much. >> >> It looks very weird. I like that. > >You can cancel the d.c. base current error (and use the faster >transistor without fear) if you tack on a bunch of resistors--I worked >up a PNP version of the technique whit3rd's citing--but the Phil- >lington is cuter and mostly eliminates the need.Here's a "shorted cascode" with Ib correction. https://dl.dropbox.com/u/53724080/Circuits/Isrc_cascode_Ib.JPG It would need some work on the slow loop dynamics. -- John Larkin Highland Technology, Inc jlarkin at highlandtechnology dot com http://www.highlandtechnology.com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom laser drivers and controllers Photonics and fiberoptic TTL data links VME thermocouple, LVDT, synchro acquisition and simulation
Reply by ●August 16, 20122012-08-16
John Larkin wrote:> > On Thu, 16 Aug 2012 17:28:16 -0400, Phil Hobbs > <pcdhSpamMeSenseless@electrooptical.net> wrote: > > >On 08/16/2012 05:19 PM, John Larkin wrote: > >> On Thu, 16 Aug 2012 12:20:49 -0700 (PDT), dagmargoodboat@yahoo.com > >> wrote: > >> > >>> On Aug 16, 11:29 am, Jim Thompson<To-Email-Use-The-Envelope-I...@On- > >>> My-Web-Site.com> wrote: > >>>> On Thu, 16 Aug 2012 07:39:04 -0700 (PDT), dagmargoodb...@yahoo.com > >>>> wrote: > >>>> > >>>> > >>>> > >>>> > >>>> > >>>> > >>>> > >>>> > >>>> > >>>>> On Aug 15, 10:43 pm, John Larkin > >>>>> <jjlar...@highNOTlandTHIStechnologyPART.com> wrote: > >>>> > >>>>>> Hey, how about this? > >>>> > >>>>>> https://dl.dropbox.com/u/53724080/Circuits/Isrc_simpler.JPG > >>>> > >>>>> It's kind of a variation on Phil's active inductor. That defeats all > >>>>> the oscillation modes I can think of so far. (c-b, e-b, c-e)(Which > >>>>> means little, since circuits are so ingenious inventing new ones.) > >>>> > >>>>>> Ludicrously simple. No Miller nonsense. Might not oscillate. MMBTH81 > >>>>>> would have about 2 pF out, or BFT92 would have about 0.75 but lower > >>>>>> beta. > >>>> > >>>>> You could use the beta-cancellation trick. But, stray and parasitic > >>>>> capacitance will be big factors compared to a 10pF integration cap. > >>>> > >>>>>> If this works, I should have done it years ago. > >>>> > >>>> Yup. But I'd still add a base resistor, for the very reasons Gerhard > >>>> Hoffmann brought up... the output Z of the OpAmp is a big unknown (*). > >>>> > >>>> (*) Which brings to mind... maybe add some DC load to the OpAmp to > >>>> keep it out of the class-B region. > >>> > >>> I agree with both those points, and meant to mention them. I don't > >>> trust the op-amp to behave nicely to impulses, or its output to be > >>> stiff. I'd expect the output to be hi-z on this timescale. > >>> > >>> If anything I'd be tempted to bypass the bjt's base to the supply > >>> rail. The BJT then handles the fast stuff. Isolate that capacitive > >>> load from the op amp's output with a series resistor, and stabilize > >>> the op amp loop itself with a feed-forward cap. Standard stuff. > >>> > >>> Does it all matter? Unclamping the integration cap will feed-thru a > >>> small impulse through to the bjt / CCS. After that, it's pretty > >>> clean. So, the op-amp's output impedance might not matter, but I > >>> suspect it will. > >>> > >>> I invented loading LM324 outputs all by myself waayyyy back as a punk > >>> kid, to stiffen them up and kill the crossover THD. Of course > >>> everyone else on the planet thought of it too. :-) > >> > >> > >> Me too! And I also invented loading the charge pump phase detector in > >> the 4046. > >> > >> I also invented the successive-detection log video detector, and the > >> dual-slope ADC. I did those at times when I could have patented them. > >> I'd be smoking seegars on the Riviera. > >> > >> > >DLVAs have been around since at least the war: e.g. US Patent 2577506 to > >Belleville, filed 1945. > > > >Cheers > > > >Phil Hobbs > > Oh well. But I did invent the dual-slope some years before Fairchild > (?) patented it. >I used to be a big fan of the Intersil ICL7109 back in the day. Dual slope was magic. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
Reply by ●August 17, 20122012-08-17
On Aug 17, 4:01=A0am, Phil Hobbs <pcdhSpamMeSensel...@electrooptical.net> wrote:> John Larkin wrote: > > > On Thu, 16 Aug 2012 17:28:16 -0400, Phil Hobbs > > <pcdhSpamMeSensel...@electrooptical.net> wrote: > > > >On 08/16/2012 05:19 PM, John Larkin wrote: > > >> On Thu, 16 Aug 2012 12:20:49 -0700 (PDT), dagmargoodb...@yahoo.com > > >> wrote: > > > >>> On Aug 16, 11:29 am, Jim Thompson<To-Email-Use-The-Envelope-I...@On=-> > >>> My-Web-Site.com> =A0wrote: > > >>>> On Thu, 16 Aug 2012 07:39:04 -0700 (PDT), dagmargoodb...@yahoo.com > > >>>> wrote: > > > >>>>> On Aug 15, 10:43 pm, John Larkin > > >>>>> <jjlar...@highNOTlandTHIStechnologyPART.com> =A0wrote: > > > >>>>>> Hey, how about this? > > > >>>>>>https://dl.dropbox.com/u/53724080/Circuits/Isrc_simpler.JPG > > > >>>>> It's kind of a variation on Phil's active inductor. =A0That defea=ts all> > >>>>> the oscillation modes I can think of so far. (c-b, e-b, c-e)(Whic=h> > >>>>> means little, since circuits are so ingenious inventing new ones.=)> > > >>>>>> Ludicrously simple. No Miller nonsense. Might not oscillate. MMB=TH81> > >>>>>> would have about 2 pF out, or BFT92 would have about 0.75 but lo=wer> > >>>>>> beta. > > > >>>>> You could use the beta-cancellation trick. =A0But, stray and para=sitic> > >>>>> capacitance will be big factors compared to a 10pF integration ca=p.> > > >>>>>> If this works, I should have done it years ago. > > > >>>> Yup. =A0But I'd still add a base resistor, for the very reasons Ge=rhard> > >>>> Hoffmann brought up... the output Z of the OpAmp is a big unknown =(*).> > > >>>> (*) Which brings to mind... maybe add some DC load to the OpAmp to > > >>>> keep it out of the class-B region. > > > >>> I agree with both those points, and meant to mention them. =A0I don='t> > >>> trust the op-amp to behave nicely to impulses, or its output to be > > >>> stiff. =A0I'd expect the output to be hi-z on this timescale. > > > >>> If anything I'd be tempted to bypass the bjt's base to the supply > > >>> rail. The BJT then handles the fast stuff. =A0Isolate that capaciti=ve> > >>> load from the op amp's output with a series resistor, and stabilize > > >>> the op amp loop itself with a feed-forward cap. =A0Standard stuff. > > > >>> Does it all matter? =A0Unclamping the integration cap will feed-thr=u a> > >>> small impulse through to the bjt / CCS. =A0After that, it's pretty > > >>> clean. =A0So, the op-amp's output impedance might not matter, but I > > >>> suspect it will. > > > >>> I invented loading LM324 outputs all by myself waayyyy back as a pu=nk> > >>> kid, to stiffen them up and kill the crossover THD. =A0Of course > > >>> everyone else on the planet thought of it too. :-) > > > >> Me too! And I also invented loading the charge pump phase detector i=n> > >> the 4046. > > > >> I also invented the successive-detection log video detector, and the > > >> dual-slope ADC. I did those at times when I could have patented them=.> > >> I'd be smoking seegars on the Riviera. > > > >DLVAs have been around since at least the war: e.g. US Patent 2577506 =to> > >Belleville, filed 1945. > > > >Cheers > > > >Phil Hobbs > > > Oh well. But I did invent the dual-slope some years before Fairchild > > (?) patented it. > > I used to be a big fan of the Intersil ICL7109 back in the day. =A0Dual > slope was magic.Quad slope was a trifle more practical - I can't actually remember the details, but I could probably work it out if I had to. As usual engineers dumped elegance in favour of something messier which works better, but is harder to explain to undergraduates. http://www.analog.com/static/imported-files/tutorials/MT-027.pdf -- Bill Sloman, Nijmegen
Reply by ●August 17, 20122012-08-17
On Thu, 16 Aug 2012 16:39:55 -0700, John Larkin wrote:> On Thu, 16 Aug 2012 15:08:21 -0700 (PDT), dagmargoodboat@yahoo.com > wrote: > >>On Aug 16, 5:15 pm, John Larkin <jlar...@highlandtechnology.com> wrote: >>> On Thu, 16 Aug 2012 13:12:50 -0700 (PDT), dagmargoodb...@yahoo.com >>> wrote: >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> >On Aug 16, 3:59 pm, Jim Thompson >>> ><To-Email-Use-The-Envelope-I...@On-My- Web-Site.com> wrote: >>> >> On Thu, 16 Aug 2012 12:24:51 -0700 (PDT), dagmargoodb...@yahoo.com >>> >> wrote: >>> >>> >> >On Aug 16, 11:41 am, John Larkin >>> >> ><jjlar...@highNOTlandTHIStechnologyPART.com> wrote: >>> >> >> On Thu, 16 Aug 2012 08:29:46 -0700, Jim Thompson >>> >>> >> >> <To-Email-Use-The-Envelope-I...@On-My-Web-Site.com> wrote: >>> >> >> >On Thu, 16 Aug 2012 07:39:04 -0700 (PDT), >>> >> >> >dagmargoodb...@yahoo.com wrote: >>> >>> >> >> >>On Aug 15, 10:43 pm, John Larkin >>> >> >> >><jjlar...@highNOTlandTHIStechnologyPART.com> wrote: >>> >>> >> >> >>> Hey, how about this? >>> >>> >> >> >>>https://dl.dropbox.com/u/53724080/Circuits/Isrc_simpler.JPG >>> >>> >> >> >>It's kind of a variation on Phil's active inductor. That >>> >> >> >>defeats all the oscillation modes I can think of so far. (c-b, >>> >> >> >>e-b, c-e)(Which means little, since circuits are so ingenious >>> >> >> >>inventing new ones.) >>> >>> >> >> >>> Ludicrously simple. No Miller nonsense. Might not oscillate. >>> >> >> >>> MMBTH81 would have about 2 pF out, or BFT92 would have about >>> >> >> >>> 0.75 but lower beta. >>> >>> >> >> >>You could use the beta-cancellation trick. But, stray and >>> >> >> >>parasitic capacitance will be big factors compared to a 10pF >>> >> >> >>integration cap. >>> >>> >> >> >>> If this works, I should have done it years ago. >>> >>> >> >> >Yup. But I'd still add a base resistor, for the very reasons >>> >> >> >Gerhard Hoffmann brought up... the output Z of the OpAmp is a >>> >> >> >big unknown (*). >>> >>> >> >> The base and emitter are shorted at AC. The opamp Zout doesn't >>> >> >> matter any more. >>> >>> >> >Yep, that's a good point. If the op amp can drive that load all >>> >> >by itself, it seems okay. That load being 1nF in series with >>> >> >120ohms, it should be a snap. >>> >>> >> >> >(*) Which brings to mind... maybe add some DC load to the OpAmp >>> >> >> >to keep it out of the class-B region. >>> >>> >> >> LM7301 is a r-r output design. It isn't an ancient class-B like >>> >> >> an LM324. >>> >>> >> >I love LM324As. Yeah, I know you hate 'em. >>> >>> >> CMOS R-R OpAmps inherently have load-dependent gain and phase, so I >>> >> avoid them when I can. >>> >>> >But John rightly points out that in the ramp timescale, Vbe is held >>> >constant by the 1nF cap. So, the transistor's fine. >>> >>> >In the longer term a load-step on the op amp's output may make it >>> >overshoot or ring. That would happen opening and closing the >>> >shorting switch. Probably not a big deal, AFAICT. >>> >>> >I was initially concerned the op amp is effectively absent on the 1nS >>> >response level, so there's no base drive, but John's right, that >>> >behavior is set by the 1nF b-e cap. IOW, it doesn't matter. >>> >>> The opamp is slowly servoing Vbe to get the average current right. The >>> ramp is so fast that the transistor is still running in constant Vbe >>> mode during the ramp. The cap could actually be much bigger... doesn't >>> matter much. >>> >>> It looks very weird. I like that. >> >>You can cancel the d.c. base current error (and use the faster >>transistor without fear) if you tack on a bunch of resistors--I worked >>up a PNP version of the technique whit3rd's citing--but the Phil- >>lington is cuter and mostly eliminates the need. > > > Here's a "shorted cascode" with Ib correction. > > https://dl.dropbox.com/u/53724080/Circuits/Isrc_cascode_Ib.JPG > > It would need some work on the slow loop dynamics.Very cute! This configuration may not be the best if you've got ripple on the positive supply, but other than that (and the whole Ccb nonlinearity problem, which none of the configurations address) it looks promising! In fact, if you reconnect the ground lead of the lowest capacitor to the positive supply, the whole thing is referenced to the positive supply which should reduce any ripple problems. Oscillation - that's another problem, you'll want to have a low HF impedance on the V+ supply anyway. -F
Reply by ●August 17, 20122012-08-17
On Fri, 17 Aug 2012 15:07:40 +0000 (UTC), Frank Miles <fpm@u.washington.edu> wrote:>On Thu, 16 Aug 2012 16:39:55 -0700, John Larkin wrote: > >> On Thu, 16 Aug 2012 15:08:21 -0700 (PDT), dagmargoodboat@yahoo.com >> wrote: >> >>>On Aug 16, 5:15�pm, John Larkin <jlar...@highlandtechnology.com> wrote: >>>> On Thu, 16 Aug 2012 13:12:50 -0700 (PDT), dagmargoodb...@yahoo.com >>>> wrote: >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >On Aug 16, 3:59�pm, Jim Thompson >>>> ><To-Email-Use-The-Envelope-I...@On-My- Web-Site.com> wrote: >>>> >> On Thu, 16 Aug 2012 12:24:51 -0700 (PDT), dagmargoodb...@yahoo.com >>>> >> wrote: >>>> >>>> >> >On Aug 16, 11:41 am, John Larkin >>>> >> ><jjlar...@highNOTlandTHIStechnologyPART.com> wrote: >>>> >> >> On Thu, 16 Aug 2012 08:29:46 -0700, Jim Thompson >>>> >>>> >> >> <To-Email-Use-The-Envelope-I...@On-My-Web-Site.com> wrote: >>>> >> >> >On Thu, 16 Aug 2012 07:39:04 -0700 (PDT), >>>> >> >> >dagmargoodb...@yahoo.com wrote: >>>> >>>> >> >> >>On Aug 15, 10:43 pm, John Larkin >>>> >> >> >><jjlar...@highNOTlandTHIStechnologyPART.com> wrote: >>>> >>>> >> >> >>> Hey, how about this? >>>> >>>> >> >> >>>https://dl.dropbox.com/u/53724080/Circuits/Isrc_simpler.JPG >>>> >>>> >> >> >>It's kind of a variation on Phil's active inductor. That >>>> >> >> >>defeats all the oscillation modes I can think of so far. (c-b, >>>> >> >> >>e-b, c-e)(Which means little, since circuits are so ingenious >>>> >> >> >>inventing new ones.) >>>> >>>> >> >> >>> Ludicrously simple. No Miller nonsense. Might not oscillate. >>>> >> >> >>> MMBTH81 would have about 2 pF out, or BFT92 would have about >>>> >> >> >>> 0.75 but lower beta. >>>> >>>> >> >> >>You could use the beta-cancellation trick. But, stray and >>>> >> >> >>parasitic capacitance will be big factors compared to a 10pF >>>> >> >> >>integration cap. >>>> >>>> >> >> >>> If this works, I should have done it years ago. >>>> >>>> >> >> >Yup. But I'd still add a base resistor, for the very reasons >>>> >> >> >Gerhard Hoffmann brought up... the output Z of the OpAmp is a >>>> >> >> >big unknown (*). >>>> >>>> >> >> The base and emitter are shorted at AC. The opamp Zout doesn't >>>> >> >> matter any more. >>>> >>>> >> >Yep, that's a good point. �If the op amp can drive that load all >>>> >> >by itself, it seems okay. �That load being 1nF in series with >>>> >> >120ohms, it should be a snap. >>>> >>>> >> >> >(*) Which brings to mind... maybe add some DC load to the OpAmp >>>> >> >> >to keep it out of the class-B region. >>>> >>>> >> >> LM7301 is a r-r output design. It isn't an ancient class-B like >>>> >> >> an LM324. >>>> >>>> >> >I love LM324As. �Yeah, I know you hate 'em. >>>> >>>> >> CMOS R-R OpAmps inherently have load-dependent gain and phase, so I >>>> >> avoid them when I can. >>>> >>>> >But John rightly points out that in the ramp timescale, Vbe is held >>>> >constant by the 1nF cap. �So, the transistor's fine. >>>> >>>> >In the longer term a load-step on the op amp's output may make it >>>> >overshoot or ring. �That would happen opening and closing the >>>> >shorting switch. �Probably not a big deal, AFAICT. >>>> >>>> >I was initially concerned the op amp is effectively absent on the 1nS >>>> >response level, so there's no base drive, but John's right, that >>>> >behavior is set by the 1nF b-e cap. �IOW, it doesn't matter. >>>> >>>> The opamp is slowly servoing Vbe to get the average current right. The >>>> ramp is so fast that the transistor is still running in constant Vbe >>>> mode during the ramp. The cap could actually be much bigger... doesn't >>>> matter much. >>>> >>>> It looks very weird. I like that. >>> >>>You can cancel the d.c. base current error (and use the faster >>>transistor without fear) if you tack on a bunch of resistors--I worked >>>up a PNP version of the technique whit3rd's citing--but the Phil- >>>lington is cuter and mostly eliminates the need. >> >> >> Here's a "shorted cascode" with Ib correction. >> >> https://dl.dropbox.com/u/53724080/Circuits/Isrc_cascode_Ib.JPG >> >> It would need some work on the slow loop dynamics. > >Very cute! This configuration may not be the best if you've got ripple >on the positive supply, but other than that (and the whole Ccb >nonlinearity problem, which none of the configurations address) it looks >promising! In fact, if you reconnect the ground lead of the lowest >capacitor to the positive supply, the whole thing is referenced to the >positive supply which should reduce any ripple problems. Oscillation - >that's another problem, you'll want to have a low HF impedance on the V+ >supply anyway. > > -FThe killer is probably the Early slope of the PNP, which will make the current source look resistive. Since RF parts are not specified well for DC behavior, one would have to measure some transistors to see how bad that may be. I think the LPTM said that SiGe transistors have huge Early voltages, but they probably don't come in PNP. Still, the Ib correction is cute, and its loop compensation turns out to be interesting. Perhaps some Master Circuit Designer will address that issue. We could dump the b-e cap to fix the Early problem and add a base resistor to kill RF oscillation in the PNP. Something like this: https://dl.dropbox.com/u/53724080/Circuits/Isrc_Ib_2.JPG As you note, C1 might better return to V+. V+ would be a bypassed copper pour, pretty stiff from DC to daylight. Pity, I really liked the Shorted Cascode. -- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom timing and laser controllers Photonics and fiberoptic TTL data links VME analog, thermocouple, LVDT, synchro, tachometer Multichannel arbitrary waveform generators
